```json { "table_name": "analysis_runs", "primary_key": "jobs_id", "candidates": [ { "rank": 0, "partition_key": "jobs_id", "partition_key_label": "primary_identifier", "partition_key_reasoning": "Perfect uniqueness (RPUV=1.00, 116 unique values for 116 rows), monotonic (1.000), serves as the primary identifier for analysis run records. Semantic role: unique job identifier.", "clustering_key": null, "clustering_key_label": null, "clustering_key_reasoning": "Dimension tables typically do not require internal sorting. No temporal or hierarchical ordering needed.", "world_line": null, "session_column": null, "confidence": "HIGH", "notes": "This is a dimension table tracking analysis run metadata. jobs_id is the natural primary key with perfect uniqueness. internal_run_id is also unique (RPUV=1.00) and could serve as an alternative identifier, but jobs_id is the integer surrogate key and more appropriate for partitioning. Many columns are 100% null (report_id, intent, code_version, dataset_version_hash, analysis_config_json, error_text), suggesting this table captures run-level facts with optional attributes. The table links to jobs table via jobs_id." } ] } ```